Ivermectin Doesn't Work For COVID-19
The end to a long and complicated story
Declaration: I have never received funding from any pharmaceutical company. All of my funding comes from the Australian state and federal government, largely through grants (although I do sometimes apply for grants from other charitable foundations). I have always publicly noted that it would be brilliant if ivermectin worked, as it’s a very safe and cheap medication, but have tried to follow the data for this important question.
If you’re not already familiar with it, ivermectin is a wonderfully effective anti-parasitic medication that people thought might work against COVID-19. It’s very safe when given in the low doses needed to treat parasites, and extremely cheap, making people quite excited about its potential as a COVID-19 treatment. I’ve been writing about ivermectin for COVID-19 since July 2021, when myself and a group of colleagues discovered that a large portion of the research underlying the drug had almost certainly been fabricated.
At that point, in 2021, we were looking at a small pool of research, and so one or two big studies being problematic completely changed the picture. When several Iranian, Egyptian, and South American papers turned out to be seriously dodgy, it entirely shifted the story of ivermectin from a pill that might work quite well to one that looked a lot less likely to be useful for the pandemic.
However, it’s been nearly three years from those days, and we now have a huge quantity of data on the question. The final large randomized trial of the bunch just came out, and it reinforces the very solid finding that ivermectin has no benefit when used in the treatment of COVID-19.
PRINCIPLE
The final large randomized trial on ivermectin that was being run is called PRINCIPLE. It’s an independently-funded platform trial that has been going in the UK for several years, testing a variety of medications for use in the pandemic. The trial originally intended to randomized about 10,000 people, testing half a dozen medications, but was stopped early for futility after including 8,811.
Platform trials are a really interesting type of scientific research, so forgive me while I briefly nerd out. In most clinical trials, you do the difficult but fundamentally simple thing where you get a bunch of hospitals and clinics to randomize groups of people to either two or more medications, and then collect the same data on those people over time to see if anything changes. In a platform trial, you instead have a multilayered protocol whereby you can introduce new drugs into the trial and stop others if they don’t work as you go. Rather than the fixed, inflexible protocol that we usually see, they have complex, malleable designs that allow the researchers to move with the times. This has proven incredibly important during the pandemic - most of the biggest, most impactful randomized trials in the last 4 years have been some form of platform trial.
In this case, the authors were testing whether different medications would work against COVID-19 in people who had relatively mild disease. At the start of the trial, this included older people or those with comorbidities, but the inclusion criteria were expanded during the trial to include everyone over the age of 18 with test-positive COVID-19. For ivermectin, PRINCIPLE tested a standard dose similar to the one that was used in the positive trials of 2020 and 2021, which worked out to about 0.2-3 mg/kg/day. This is still well above the dose that was initially believed to work, but below the higher doses that some proponents of the drug recommend.
The study had a really complex methodology and statistical analysis plan, which I recommend reading if you’re into clinical trial design, but basically it was an impressively robust trial. There was, however, one thing about the trial that was potentially a bit weak - the authors decided for pragmatic reasons to run the trial open-label. This means that there was no blinding - everyone knew which drug they were taking.
The Results
The study used two main primary outcomes - self-reported symptom alleviation, and hospitalization/death. The authors found that the average time it took people to report being recovered from COVID-19 was 16 days, but for the ivermectin group this was reduced to 14 days, with a very statistically significant result. This finding was consistent across most of the symptoms reported, and there was even a slight reduction in Long COVID reporting in the ivermectin group as well. The authors described these results as “clinically insignificant”, as they did not meet the pre-specified criteria in the trial for clinical importance.
However, for the more objective outcomes of hospitalization and death, the story was very different. The study found that 1.5% of the people who were randomized to a placebo pill at the time of the ivermectin arm of the study went to hospital or died, compared to 1.6% of people who received ivermectin*.
If you look at these results simply, it seems like ivermectin might have a benefit, even if very small. But remember - the study was open label. So there was a slight benefit for people who took ivermectin and knew they were taking it when compared to people who were told they were given a placebo on self-reported outcomes. Conversely, on objective indicators like being sick enough to die, ivermectin had no benefit. This is what we’d usually describe as reporting bias - people who know that they are taking a medication think that they’ve recovered a bit faster, but there’s no benefit on objective measures that can’t be influenced by belief.
The final results are pretty much impossible to explain if ivermectin has a benefit for COVID-19, but fit very well into the data that already shows it probably doesn’t. I’ve been putting the clinical trial results for ivermectin papers that weren’t fraudulent or seriously flawed into a meta-analysis model for a while now - here’s what the results show when you include or don’t include PRINCIPLE:
These are called forest plots, and they show the numbers from all of the included studies when you put them into a random-effects meta-analysis in Stata. You can see the aggregated finding in the little diamond at the bottom - with PRINCIPLE, the overall impact is 0.87 (0.68-1.13). In total, these papers have studied well over 10,000 people in robust randomized trials. With or without PRINCIPLE, the results are essentially identical. A very technical statistical interpretation of these numbers is that we are still uncertain of the impact of ivermectin, and the confidence interval includes both modest benefit and harm, but from a clinical perspective the evidence is now quite clear that ivermectin doesn’t work in the treatment of COVID-19.
The End Of The Story
After a wild four years, this really is the final word. We now have four huge, independently-funded randomized trials with transparent methodology and the highest standards of scientific conduct that did not find a benefit for ivermectin. There are a number of smaller and less rigorous papers that have also failed to show any impact. The only studies which seemed to indicate a benefit for the drug were filled with errors - once you discard the fabricated papers and the ones with basic mistakes, there is a clear answer to the question of ivermectin for COVID-19.
At this point, it would be unethical to start a new trial of the drug. In fact, PRINCIPLE had a pre-specified stopping point for futility. This is when the researchers set up an independent committee before they start their trial who reviews the results at fixed intervals. If the results are not showing evidence of a benefit, and are unlikely to at any point using a predetermined statistical formula, the trial is stopped so that no more patients are given a treatment that we already know to be ineffective. PRINCIPLE was stopped early using this transparent protocol, because ivermectin was clearly showing no benefit for hospitalization or death, and the committee decided that it was unethical to keep giving people ivermectin.
That means that this is the end of the line - there is not now nor will there likely ever be more evidence on ivermectin for COVID-19. The chances that it has any use as a treatment are infinitesimally small. We’ve failed to see any improvement when people are given the drug early or late, when they have severe disease or are more mild cases, when they take it on a full stomach or empty, when they take it with or without other drugs, vaccinated or not, or in any other situation that people have raised as a reason for the trials to have failed.
I’m sure this will not convince everyone. Ivermectin has taken on a life of its own in the discussion about COVID-19. For people who pinned their hopes on ivermectin as a wonder drug, and those who are now making vast sums promoting and selling ivermectin as a treatment for literally all disease, it’s hard to imagine any change of heart. It’s difficult to convince people that of the truth when they’ve staked their income, professional reputation, and even political hopes on a drug working.
But for the vast majority of us, the question has been answered. Ivermectin is a brilliant medication for parasitic infections, but it just doesn’t work for COVID-19.
*Note: if you read the study, this is presented a bit confusingly because the authors also reported the number of hospitalizations/deaths that occurred in the PRE-ivermectin part of the trial. However, this does not form part of their main analysis, as it includes patients who were part of the study in 2020, and had very different inclusion criteria and disease than the ivermectin arm.
Thanks for a clear and complete review. But.. your definition of stopping a trial for futility isn't quite right. "Futility" doesn't mean the data demonstratenthat the drug doesn't work, just that the current trial (if it were completed) has little chance of showing that the treatment does work. Reviewed here: https://evidence.nejm.org/doi/abs/10.1056/EVIDctw2100020#:~:text=Abstract,low%20to%20support%20meaningful%20comparisons.